Event Driven Service Composition for Pervasive Computing

Pervasive Computing plays an integral role in our daily life. As pervasive services become increasingly important to be able to capture context, it is reasonable to speculate that dynamic representations for composition of services is very much needed. Such representations may typically include activities or functional modules that are not only temporal but also responsive to each other. Such actions may be associated with actors who execute them, spatial and temporal information, information-flow and background objects. Consequently, one may model an activity network that depicts causal relations between activities and has the capability of embedding the semantic relations of these activities with other contextual entities. In this paper we discuss a new way of representing such dynamic workflows with a novel data structure called Activity Logic Graphs (ALGs). We introduce a novel design of an architecture that grasps the semantics of activities (or events) in dynamic pervasive environments, maps that to a collaboration of services (as ALGs) in order to recognize specific situations (known as Situation Boundaries) that has risen out of the events, and implements the recognition process using an efficient indexing technique. We test the model with a set of ALGs and report experimental results on the queries that are essential for service composition.